ResourceManager.h

Go to the documentation of this file.

//===--------------------- ResourceManager.h --------------------*- C++ -*-===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
/// \file
///
/// The classes here represent processor resource units and their management
/// strategy.  These classes are managed by the Scheduler.
///
//===----------------------------------------------------------------------===//
 
#ifndef LLVM_MCA_HARDWAREUNITS_RESOURCEMANAGER_H
#define LLVM_MCA_HARDWAREUNITS_RESOURCEMANAGER_H
 
#include "llvm/ADT/DenseMap.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/MC/MCSchedule.h"
#include "llvm/MCA/Instruction.h"
#include "llvm/MCA/Support.h"
#include "llvm/Support/Compiler.h"
 
namespace llvm {
namespace mca {
 
/// Used to notify the internal state of a processor resource.
///
/// A processor resource is available if it is not reserved, and there are
/// available slots in the buffer.  A processor resource is unavailable if it
/// is either reserved, or the associated buffer is full. A processor resource
/// with a buffer size of -1 is always available if it is not reserved.
///
/// Values of type ResourceStateEvent are returned by method
/// ResourceManager::canBeDispatched()
///
/// The naming convention for resource state events is:
///  * Event names start with prefix RS_
///  * Prefix RS_ is followed by a string describing the actual resource state.
enum ResourceStateEvent {
  RS_BUFFER_AVAILABLE,
  RS_BUFFER_UNAVAILABLE,
  RS_RESERVED
};
 
/// Resource allocation strategy used by hardware scheduler resources.
class LLVM_ABI ResourceStrategy {
  ResourceStrategy(const ResourceStrategy &) = delete;
  ResourceStrategy &operator=(const ResourceStrategy &) = delete;
 
public:
  ResourceStrategy() = default;
  virtual ~ResourceStrategy();
 
  /// Selects a processor resource unit from a ReadyMask.
  virtual uint64_t select(uint64_t ReadyMask) = 0;
 
  /// Called by the ResourceManager when a processor resource group, or a
  /// processor resource with multiple units has become unavailable.
  ///
  /// The default strategy uses this information to bias its selection logic.
  virtual void used(uint64_t ResourceMask) {}
};
 
/// Default resource allocation strategy used by processor resource groups and
/// processor resources with multiple units.
class LLVM_ABI DefaultResourceStrategy final : public ResourceStrategy {
  /// A Mask of resource unit identifiers.
  ///
  /// There is one bit set for every available resource unit.
  /// It defaults to the value of field ResourceSizeMask in ResourceState.
  const uint64_t ResourceUnitMask;
 
  /// A simple round-robin selector for processor resource units.
  /// Each bit of this mask identifies a sub resource within a group.
  ///
  /// As an example, lets assume that this is a default policy for a
  /// processor resource group composed by the following three units:
  ///   ResourceA -- 0b001
  ///   ResourceB -- 0b010
  ///   ResourceC -- 0b100
  ///
  /// Field NextInSequenceMask is used to select the next unit from the set of
  /// resource units. It defaults to the value of field `ResourceUnitMasks` (in
  /// this example, it defaults to mask '0b111').
  ///
  /// The round-robin selector would firstly select 'ResourceC', then
  /// 'ResourceB', and eventually 'ResourceA'.  When a resource R is used, the
  /// corresponding bit in NextInSequenceMask is cleared.  For example, if
  /// 'ResourceC' is selected, then the new value of NextInSequenceMask becomes
  /// 0xb011.
  ///
  /// When NextInSequenceMask becomes zero, it is automatically reset to the
  /// default value (i.e. ResourceUnitMask).
  uint64_t NextInSequenceMask;
 
  /// This field is used to track resource units that are used (i.e. selected)
  /// by other groups other than the one associated with this strategy object.
  ///
  /// In LLVM processor resource groups are allowed to partially (or fully)
  /// overlap. That means, a same unit may be visible to multiple groups.
  /// This field keeps track of uses that have originated from outside of
  /// this group. The idea is to bias the selection strategy, so that resources
  /// that haven't been used by other groups get prioritized.
  ///
  /// The end goal is to (try to) keep the resource distribution as much uniform
  /// as possible. By construction, this mask only tracks one-level of resource
  /// usage. Therefore, this strategy is expected to be less accurate when same
  /// units are used multiple times by other groups within a single round of
  /// select.
  ///
  /// Note: an LRU selector would have a better accuracy at the cost of being
  /// slightly more expensive (mostly in terms of runtime cost). Methods
  /// 'select' and 'used', are always in the hot execution path of llvm-mca.
  /// Therefore, a slow implementation of 'select' would have a negative impact
  /// on the overall performance of the tool.
  uint64_t RemovedFromNextInSequence;
 
public:
  DefaultResourceStrategy(uint64_t UnitMask)
      : ResourceUnitMask(UnitMask), NextInSequenceMask(UnitMask),
        RemovedFromNextInSequence(0) {}
  ~DefaultResourceStrategy() override = default;
 
  uint64_t select(uint64_t ReadyMask) override;
  void used(uint64_t Mask) override;
};
 
/// A processor resource descriptor.
///
/// There is an instance of this class for every processor resource defined by
/// the machine scheduling model.
/// Objects of class ResourceState dynamically track the usage of processor
/// resource units.
class ResourceState {
  /// An index to the MCProcResourceDesc entry in the processor model.
  const unsigned ProcResourceDescIndex;
 
  /// A resource mask. This is generated by the tool with the help of
  /// function `mca::computeProcResourceMasks' (see Support.h).
  ///
  /// Field ResourceMask only has one bit set if this resource state describes a
  /// processor resource unit (i.e. this is not a group). That means, we can
  /// quickly check if a resource is a group by simply counting the number of
  /// bits that are set in the mask.
  ///
  /// The most significant bit of a mask (MSB) uniquely identifies a resource.
  /// Remaining bits are used to describe the composition of a group (Group).
  ///
  /// Example (little endian):
  ///            Resource |  Mask      |  MSB       |  Group
  ///            ---------+------------+------------+------------
  ///            A        |  0b000001  |  0b000001  |  0b000000
  ///                     |            |            |
  ///            B        |  0b000010  |  0b000010  |  0b000000
  ///                     |            |            |
  ///            C        |  0b010000  |  0b010000  |  0b000000
  ///                     |            |            |
  ///            D        |  0b110010  |  0b100000  |  0b010010
  ///
  /// In this example, resources A, B and C are processor resource units.
  /// Only resource D is a group resource, and it contains resources B and C.
  /// That is because MSB(B) and MSB(C) are both contained within Group(D).
  const uint64_t ResourceMask;
 
  /// This field is set iff this resource is a group resource. This class can
  /// represent either a processor resource unit or a processor resource group
  /// composition.
  const bool IsAGroup;
 
  /// A ProcResource can have multiple units.
  ///
  /// For processor resource groups this field is a mask of contained resource
  /// units. It is obtained from ResourceMask by clearing the highest set bit.
  /// The number of resource units in a group can be simply computed as the
  /// population count of this field.
  ///
  /// For normal (i.e. non-group) resources, the number of bits set in this mask
  /// is equivalent to the number of units declared by the processor model (see
  /// field 'NumUnits' in 'ProcResourceUnits').
  const uint64_t ResourceSizeMask;
 
  /// A mask of ready units.
  uint64_t ReadyMask;
 
  /// Buffered resources will have this field set to a positive number different
  /// than zero. A buffered resource behaves like a reservation station
  /// implementing its own buffer for out-of-order execution.
  ///
  /// A BufferSize of 1 is used by scheduler resources that force in-order
  /// execution.
  ///
  /// A BufferSize of 0 is used to model in-order issue/dispatch resources.
  /// Since in-order issue/dispatch resources don't implement buffers, dispatch
  /// events coincide with issue events.
  /// Also, no other instruction ca be dispatched/issue while this resource is
  /// in use. Only when all the "resource cycles" are consumed (after the issue
  /// event), a new instruction ca be dispatched.
  const int BufferSize;
 
  /// Available slots in the buffer (zero, if this is not a buffered resource).
  unsigned AvailableSlots;
 
  /// This field is set if this resource is currently reserved.
  ///
  /// Resources can be reserved for a number of cycles.
  /// Instructions can still be dispatched to reserved resources. However,
  /// istructions dispatched to a reserved resource cannot be issued to the
  /// underlying units (i.e. pipelines) until the resource is released.
  bool Unavailable;
 
  /// Checks for the availability of unit 'SubResMask' in the group.
  bool isSubResourceReady(uint64_t SubResMask) const {
    return ReadyMask & SubResMask;
  }
 
public:
  LLVM_ABI ResourceState(const MCProcResourceDesc &Desc, unsigned Index,
                         uint64_t Mask, const MCSchedModel &SM,
                         float ReservationStationScaleFactor);
 
  unsigned getProcResourceID() const { return ProcResourceDescIndex; }
  uint64_t getResourceMask() const { return ResourceMask; }
  uint64_t getReadyMask() const { return ReadyMask; }
  int getBufferSize() const { return BufferSize; }
 
  bool isBuffered() const { return BufferSize > 0; }
  bool isInOrder() const { return BufferSize == 1; }
 
  /// Returns true if this is an in-order dispatch/issue resource.
  bool isADispatchHazard() const { return BufferSize == 0; }
  bool isReserved() const { return Unavailable; }
 
  void setReserved() { Unavailable = true; }
  void clearReserved() { Unavailable = false; }
 
  /// Returs true if this resource is not reserved, and if there are at least
  /// `NumUnits` available units.
  LLVM_ABI bool isReady(unsigned NumUnits = 1) const;
 
  uint64_t getNumReadyUnits() const { return llvm::popcount(ReadyMask); }
 
  bool isAResourceGroup() const { return IsAGroup; }
 
  bool containsResource(uint64_t ID) const { return ResourceMask & ID; }
 
  void markSubResourceAsUsed(uint64_t ID) {
    assert(isSubResourceReady(ID));
    ReadyMask ^= ID;
  }
 
  void releaseSubResource(uint64_t ID) {
    assert(!isSubResourceReady(ID));
    ReadyMask ^= ID;
  }
 
  unsigned getNumUnits() const {
    return isAResourceGroup() ? 1U : llvm::popcount(ResourceSizeMask);
  }
 
  /// Checks if there is an available slot in the resource buffer.
  ///
  /// Returns RS_BUFFER_AVAILABLE if this is not a buffered resource, or if
  /// there is a slot available.
  ///
  /// Returns RS_RESERVED if this buffered resource is a dispatch hazard, and it
  /// is reserved.
  ///
  /// Returns RS_BUFFER_UNAVAILABLE if there are no available slots.
  LLVM_ABI ResourceStateEvent isBufferAvailable() const;
 
  /// Reserve a buffer slot.
  ///
  /// Returns true if the buffer is not full.
  /// It always returns true if BufferSize is set to zero.
  bool reserveBuffer() {
    if (BufferSize <= 0)
      return true;
 
    --AvailableSlots;
    assert(AvailableSlots <= static_cast<unsigned>(BufferSize));
    return AvailableSlots;
  }
 
  /// Releases a slot in the buffer.
  void releaseBuffer() {
    // Ignore dispatch hazards or invalid buffer sizes.
    if (BufferSize <= 0)
      return;
 
    ++AvailableSlots;
    assert(AvailableSlots <= static_cast<unsigned>(BufferSize));
  }
 
#ifndef NDEBUG
  void dump() const;
#endif
};
 
/// A resource unit identifier.
///
/// This is used to identify a specific processor resource unit using a pair
/// of indices where the 'first' index is a processor resource mask, and the
/// 'second' index is an index for a "sub-resource" (i.e. unit).
typedef std::pair<uint64_t, uint64_t> ResourceRef;
 
// First: a MCProcResourceDesc index identifying a buffered resource.
// Second: max number of buffer entries used in this resource.
typedef std::pair<unsigned, unsigned> BufferUsageEntry;
 
/// A resource manager for processor resource units and groups.
///
/// This class owns all the ResourceState objects, and it is responsible for
/// acting on requests from a Scheduler by updating the internal state of
/// ResourceState objects.
/// This class doesn't know about instruction itineraries and functional units.
/// In future, it can be extended to support itineraries too through the same
/// public interface.
class ResourceManager {
  // Set of resources available on the subtarget.
  //
  // There is an instance of ResourceState for every resource declared by the
  // target scheduling model.
  //
  // Elements of this vector are ordered by resource kind. In particular,
  // resource units take precedence over resource groups.
  //
  // The index of a processor resource in this vector depends on the value of
  // its mask (see the description of field ResourceState::ResourceMask).  In
  // particular, it is computed as the position of the most significant bit set
  // (MSB) in the mask plus one (since we want to ignore the invalid resource
  // descriptor at index zero).
  //
  // Example (little endian):
  //
  //             Resource | Mask    |  MSB    | Index
  //             ---------+---------+---------+-------
  //                 A    | 0b00001 | 0b00001 |   1
  //                      |         |         |
  //                 B    | 0b00100 | 0b00100 |   3
  //                      |         |         |
  //                 C    | 0b10010 | 0b10000 |   5
  //
  //
  // The same index is also used to address elements within vector `Strategies`
  // and vector `Resource2Groups`.
  std::vector<std::unique_ptr<ResourceState>> Resources;
  std::vector<std::unique_ptr<ResourceStrategy>> Strategies;
 
  // Used to quickly identify groups that own a particular resource unit.
  std::vector<uint64_t> Resource2Groups;
 
  // A table that maps processor resource IDs to processor resource masks.
  SmallVector<uint64_t, 8> ProcResID2Mask;
 
  // A table that maps resource indices to actual processor resource IDs in the
  // scheduling model.
  SmallVector<unsigned, 8> ResIndex2ProcResID;
 
  // Keeps track of which resources are busy, and how many cycles are left
  // before those become usable again.
  SmallDenseMap<ResourceRef, unsigned> BusyResources;
 
  // Set of processor resource units available on the target.
  uint64_t ProcResUnitMask;
 
  // Set of processor resource units that are available during this cycle.
  uint64_t AvailableProcResUnits;
 
  // Set of processor resources that are currently reserved.
  uint64_t ReservedResourceGroups;
 
  // Set of unavailable scheduler buffer resources. This is used internally to
  // speedup `canBeDispatched()` queries.
  uint64_t AvailableBuffers;
 
  // Set of dispatch hazard buffer resources that are currently unavailable.
  uint64_t ReservedBuffers;
 
  // Returns the actual resource unit that will be used.
  ResourceRef selectPipe(uint64_t ResourceID);
 
  void use(const ResourceRef &RR);
  void release(const ResourceRef &RR);
 
  unsigned getNumUnits(uint64_t ResourceID) const;
 
  // Overrides the selection strategy for the processor resource with the given
  // mask.
  LLVM_ABI void setCustomStrategyImpl(std::unique_ptr<ResourceStrategy> S,
                                      uint64_t ResourceMask);
 
public:
  LLVM_ABI ResourceManager(const MCSchedModel &SM,
                           float ReservationStationScaleFactor);
  virtual ~ResourceManager() = default;
 
  // Overrides the selection strategy for the resource at index ResourceID in
  // the MCProcResourceDesc table.
  void setCustomStrategy(std::unique_ptr<ResourceStrategy> S,
                         unsigned ResourceID) {
    assert(ResourceID < ProcResID2Mask.size() &&
           "Invalid resource index in input!");
    return setCustomStrategyImpl(std::move(S), ProcResID2Mask[ResourceID]);
  }
 
  // Returns RS_BUFFER_AVAILABLE if buffered resources are not reserved, and if
  // there are enough available slots in the buffers.
  LLVM_ABI ResourceStateEvent canBeDispatched(uint64_t ConsumedBuffers) const;
 
  // Return the processor resource identifier associated to this Mask.
  LLVM_ABI unsigned resolveResourceMask(uint64_t Mask) const;
 
  // Acquires a slot from every buffered resource in mask `ConsumedBuffers`.
  // Units that are dispatch hazards (i.e. BufferSize=0) are marked as reserved.
  LLVM_ABI void reserveBuffers(uint64_t ConsumedBuffers);
 
  // Releases a slot from every buffered resource in mask `ConsumedBuffers`.
  // ConsumedBuffers is a bitmask of previously acquired buffers (using method
  // `reserveBuffers`). Units that are dispatch hazards (i.e. BufferSize=0) are
  // not automatically unreserved by this method.
  LLVM_ABI void releaseBuffers(uint64_t ConsumedBuffers);
 
  // Reserve a processor resource. A reserved resource is not available for
  // instruction issue until it is released.
  LLVM_ABI void reserveResource(uint64_t ResourceID);
 
  // Release a previously reserved processor resource.
  LLVM_ABI void releaseResource(uint64_t ResourceID);
 
  // Returns a zero mask if resources requested by Desc are all available during
  // this cycle. It returns a non-zero mask value only if there are unavailable
  // processor resources; each bit set in the mask represents a busy processor
  // resource unit or a reserved processor resource group.
  LLVM_ABI uint64_t checkAvailability(const InstrDesc &Desc) const;
 
  uint64_t getProcResUnitMask() const { return ProcResUnitMask; }
  uint64_t getAvailableProcResUnits() const { return AvailableProcResUnits; }
 
  using ResourceWithCycles = std::pair<ResourceRef, ReleaseAtCycles>;
 
  void issueInstruction(const InstrDesc &Desc,
                        SmallVectorImpl<ResourceWithCycles> &Pipes) {
    if (Desc.HasPartiallyOverlappingGroups)
      return issueInstructionImpl(Desc, Pipes);
 
    return fastIssueInstruction(Desc, Pipes);
  }
 
  // Selects pipeline resources consumed by an instruction.
  // This method works under the assumption that used group resources don't
  // partially overlap. The logic is guaranteed to find a valid resource unit
  // schedule, no matter in which order individual uses are processed. For that
  // reason, the vector of resource uses is simply (and quickly) processed in
  // sequence. The resulting schedule is eventually stored into vector `Pipes`.
  LLVM_ABI void
  fastIssueInstruction(const InstrDesc &Desc,
                       SmallVectorImpl<ResourceWithCycles> &Pipes);
 
  // Selects pipeline resources consumed by an instruction.
  // This method works under the assumption that used resource groups may
  // partially overlap. This complicates the selection process, because the
  // order in which uses are processed matters. The logic internally prioritizes
  // groups which are more constrained than others.
  LLVM_ABI void
  issueInstructionImpl(const InstrDesc &Desc,
                       SmallVectorImpl<ResourceWithCycles> &Pipes);
 
  LLVM_ABI void cycleEvent(SmallVectorImpl<ResourceRef> &ResourcesFreed);
 
#ifndef NDEBUG
  void dump() const {
    for (const std::unique_ptr<ResourceState> &Resource : Resources)
      Resource->dump();
  }
#endif
};
} // namespace mca
} // namespace llvm
 
#endif // LLVM_MCA_HARDWAREUNITS_RESOURCEMANAGER_H